This project is concerned with characterizing and improving the delivery of pharmacologic and diagnostic agents to the central nervous system. [unreadable] [unreadable] (1) Characteristics of the transport of viral-sized molecules in CNS tissue have been investigated. Following convective-enhanced delivery (CED) of Combidex (24nm dia. MR marker) at 0.1 microliter per minute into gray matter, Vd/Vi ratios (4.1) were found to only be slightly lower than those typical of small or protein-sized molecules (4.7 to 5.2) or even of associated-adenovirus-1 (5.5), suggesting little interaction of the MR marker with the extracellular matrix and good marker characteristics. However, in white matter Vd/Vi drops to 2.3, a value suggesting that Combidex may possibly seize in some channels, leading to a pressure rise and subsequent dilatation of the tissue. This is consistent with previous observation of a reduced Vd/Vi when Gd-loaded liposomes (124nm) have been infused into white matter. New experiments have been designed to track both macro and micro distribution of large particles by quantum dots.[unreadable] [unreadable] (2) A finite element model of the transport of GDNF in the human putamen/pallidi nuclei was developed and used to determine if the differing patient responses to GDNF reported in three recent Parkinson's disease clinical trials were caused by differences in drug delivery. GDNF distribution was modeled as arising from fluxes through the interstitial space combined with binding to heparan sulfate moieties, passive microvascular clearance, and extrachoroidal fluid dilution. Trial-specific end-hole and multiport catheters were described together with estimates of associated backflow effects. Steady state and transient results were computed. Recent overlay of predicted backflow lengths on a map of catheter tip positions indicates that for many patients in these trials, substantial drug may have been lost by backflow to non-targeted areas, thus reducing the statistical power of one or more trials. Detailed follow-up with patient-specific MR images is planned to confirm this finding.[unreadable] [unreadable] (3) The potential for delivering the drug gemcitabine to brainstem and resident tumors was characterized. A distributed pharmacokinetic model of the agent, including descriptions of extra/intracellular partitioning, microvascular and metabolic clearance, and pseudo-isotropic transport, was developed to describe its delivery by direct interstitial infusion into brainstem. A model of the Magnetic Resonance (MR) tracer GdDTPA, co-infused with gemcitabine, was also developed. Simulations revealed that GdDTPA spreads farther and more quickly than the drug and thus follows a much different time course. Consequently it may only be used to track the drug in relatively homogeneous tissue if its distribution is interpreted to precede that of gemcitabine by about 2 hours. Large volume infusions were found to lead to steady state distributions of gemcitabine but not of the MR tracer. To better characterize deamination rates of gemcitabine, parent compound and deaminated metabolite are to be quantitated in infused tissue by means of SIMS (mass spectrometry).[unreadable] [unreadable] (4) A review of convective-enhanced-delivery was prepared and delivered to the neurosurgical community at the Cleveland Brain Tumor Institute. The 12-year overview focused on quantitation and imaging, and emphasized both simple delivery estimators as well as anatomically correct finite element computational techniques.[unreadable] [unreadable] (5) A monitoring paradigm was developed for the use of GdDTPA as an MR tracer in the CED of glucocerebrosidase for the treatment of neuronopathic (Type 2) Gaucher's disease. 1 mM GdDTPA infused into rodent and non-human primate brain (frontal lobe) yielded MR images that tracked 70kD radiolabeled dextran measured by quantitative autoradiography, provided the MR was thresholded using an intensity equivalent to twice the standard deviation of the background signal on the contralateral side. The Vd time series of GdDTPA in the post-infusion phase has been analyzed by a pseudo-isotropic convection-diffusion model of the brainstem (pons) to yield first estimates of the effective clearance rate of the marker in this tissue. [unreadable] [unreadable] (6) The potential for using botulinum toxin A (btxA) for the treatment of epileptic seizure was investigated. A finite element model for the distribution of this agent in the non-human primate was developed and used to identify a protocol capable of dosing, but not overflowing, the hippocampus. A CED protocol consisting of an 8 hour infusion of 350 pM btxA through a 1.3 mm diameter catheter at 1.4 ?l/min was predicted to have sufficient distributional range for a receptor Kd of 0.3 nM, with negligible loss from homogenous tissue after cessation of infusion. The total dose of this protocol is 34 ng, or 4.5 ng/kg for a 7.5 kg non-human primate. However, given a systemic LD50 on the order of 1 ng/kg, these computations suggest substantial risk if only 22% of the dose were to escape confinement to the central nervous system (e.g. via unintended placement of the catheter in a vascular or CSF space). Hence btxA CED protocols may have to be limited to ones that avoid general hippocampal delivery and more closely target the epileptic foci. In addition, recent experimental work raises the possibility of significant non-specific interactions of btxA with target neurons and hence the need to modify its role in present models.